Pilot data for CaLLab implementation of manybabies1

et_files <- paste("data/", list.files("data/", "*.hdf5"), sep = "")
log_files <- list.files("data/", "*.log", full.names = TRUE)
trial_files <- list.files("data/", "*.csv", full.names = TRUE)

get_end_times <- function(log_file) {
  data_frame(text = read_lines(log_file)) %>%
    filter(str_detect(text, "autoDraw = False")) %>%
    filter(str_detect(text, "checkerboard") | str_detect(text, "attention_getter")) %>%
    separate(text, c("End", "EXP", "message"), sep = "\t") %>%
    mutate(type = if_else(str_detect(message, "attention_getter"), 
                          "Attention", "Trial")) %>%
    select(End, type) %>%
    mutate(trial = rep(1:(n()/2), each = 2),
           End = as.numeric(End))
}

get_trial_data <- function(trial_file) {
  read_csv(trial_file) %>%
    select(Stimulus, trials.thisTrialN, attention_key.keys, trial_key.keys) %>%
    rename(trial = trials.thisTrialN, 
           Attention = attention_key.keys, 
           Trial = trial_key.keys) %>%
    mutate(trial = trial + 1) %>%
    gather(type, key, Attention, Trial) %>%
    mutate(speechType = if_else(str_detect(Stimulus, "training"), "training",
                                if_else(str_detect(Stimulus, "ADS"), "ADS", "IDS"))) %>%
    mutate(stimNum = as.numeric(if_else(str_detect(Stimulus, "training"),
                                        as.character(trial),
                                        str_extract(Stimulus, "[0-9]+"))))
      
}

get_event_data <- function(et_file) {
  
  h5read(et_file, "data_collection/events/experiment/MessageEvent") %>%
    as_data_frame() %>%
    select(time, text) %>%
    separate(text, c("stimulus", "type"), "_") %>%
    rename(type = stimulus, event = type) %>%
    mutate(trial = rleid(type)) %>%
    spread(event, time) %>%
    group_by(type) %>%
    mutate(trial = 1:n()) 
}

# 
# order_data <- read_csv("order1.csv") %>%
#   mutate(trial = 1:nrow(.)) %>%
#   rename(item = Stimulus) %>%
#   mutate(type = if_else(item == "training", "training", 
#                         if_else(str_detect(item, "IDS"), "IDS", "ADS")))

read_data <- function(et_file, trial_file, log_file) {
  
  end_times <- get_end_times(log_file)
  trial_data <- get_trial_data(trial_file)
  event_data <- get_event_data(et_file)
  
  order_data <- filter(event_data, is.na(End)) %>%
    select(-End) %>%
    left_join(end_times, by = c("type", "trial")) %>%
    bind_rows(filter(event_data, !is.na(End))) %>%
    left_join(trial_data, by = c("type", "trial")) %>%
    arrange(trial, type)
  
  data <- h5read(et_file, "data_collection/events/eyetracker/BinocularEyeSampleEvent") %>%
    as_data_frame()
  
  filter_data <- data %>%
    select(time, status,left_gaze_x, left_gaze_y, left_eye_cam_x, 
           left_eye_cam_y,
           right_gaze_x, right_gaze_y, right_eye_cam_x, 
          right_eye_cam_y)
  
  
  tag_data <- function(row) {
    filter_data %>%
      filter(time >= row$Start & time < row$End) %>%
      mutate(type = row$type, 
             trial = row$trial)
  }
  
  
  map(seq(1,nrow(order_data)), 
         function(x) tag_data(order_data[x,])) %>%
    bind_rows() %>%
    mutate(right_gaze_x = if_else(status == 2 | status == 22,
                                  as.numeric(NA), right_gaze_x),
           right_gaze_y = if_else(status == 2 | status == 22,
                                  as.numeric(NA), right_gaze_y),
           left_gaze_x = if_else(status == 20 | status == 22,
                                 as.numeric(NA), left_gaze_x),
           left_gaze_y = if_else(status == 20 | status == 22,
                                 as.numeric(NA), left_gaze_y)) %>%
    group_by(trial, type) %>%
    mutate(time = time - min(time)) %>%
    mutate(subj = str_split(et_file, "/")[[1]][2]) %>%
    left_join(order_data, by = c("type", "trial"))
}

gaze_data <- pmap(list(et_files, trial_files, log_files), read_data) %>%
  bind_rows()

durations <- gaze_data %>%
  group_by(subj, type, speechType, trial) %>%
  summarise(duration = max(time))

datatable(durations, rownames = FALSE)

sem <- function (x, na.rm = FALSE) 
{
    if (na.rm) {
        n <- sum(!is.na(x))
    }
    else {
        n <- length(x)
    }
    stats::sd(x, na.rm = na.rm)/sqrt(n)
}

durations %>%
  filter(type == "Trial") %>%
  group_by(speechType, subj, trial) %>%
  summarise(duration = mean(duration)) %>%
  summarise(duration = mean(duration)) %>%
  spread(speechType, duration) %>%
  mutate(diff = IDS - ADS) %>%
  summarise_at(vars(diff), funs(mean, sem), na.rm = T) %>%
  datatable(rownames = FALSE)

make_plots <- function(gaze_data) {
  
  print(paste0("Subj: ", gaze_data[1,"subj"]))
  
  p1 <- ggplot(gaze_data, aes(x = time)) +
    facet_grid(trial ~ type, scales = "free") +
    geom_point(aes(y = left_gaze_x), color = "red", size = .1, alpha = .5) +
    geom_point(aes(y = right_gaze_x), color = "blue", size = .1, alpha = .5) +
    ylab("X Points")
  print(p1)
  
  p2 <- ggplot(gaze_data, aes(x = time)) +
    facet_grid(trial ~ type, scales = "free") +
    geom_point(aes(y = left_gaze_y), color = "red", size = .1, alpha = .5) +
    geom_point(aes(y = right_gaze_y), color = "blue", size = .1, alpha = .5) +
    ylab("Y Points")
  print(p2)
}

gaze_data %>%
  split(.$subj) %>%
  walk(., make_plots)

## [1] "Subj: 1_manybabies1_2017_Aug_30_1012.hdf5"

## [1] "Subj: 1_manybabies1_2017_Aug_31_1700.hdf5"

## [1] "Subj: 2_manybabies1_2017_Sep_05_1024.hdf5"

## [1] "Subj: 3_manybabies1_2017_Sep_01_1159.hdf5"